The Flame Polarization Image Characteristics Of Sintering Section And Analysis Of Sintering State

Sintering is a key step in iron and steel production.Sinter grade is directly connected with the blast furnace conditions and the output of molten iron in the process of iron making,the detection of sinter quality and sintering state mainly depends on artificial experience.Due to the terrible working conditions at the tail of the sintering machine,it is difficult for the inspectors to keep continuous observation of the flame at the tail layer.There are subjective factors affected the judgment of sinter quality,which would cause deviation.The effective information about sinter quality and sintering state is contained in the flame image of tail layer section.The feature extraction and analysis of the flame image can realize the effective judgment of different sintering states.A method based on weighted guided filtering and fast adaptive fuzzy enhancement is proposed to enhance the flame image,this paper based on the atmospheric scattering model and a first-order multi-scattering method to get the section flame image degradation model,then proposed an image restoration algorithm.Combined with the guided filter,a tolerance mechanism was used to improve the transmittance of the flame region,and then the reconstructed image was obtained.Boundary connectivity and dark channel priori are used to obtain the initial saliency map.Multi-kernel Boosting algorithm based on support vector regression was used to generate auxiliary saliency map.Then saliency maps were weighted and fused to obtain the final saliency map.It laid a foundation for improving the effective information extraction precision of flame image.In order to further analyze the flame image characteristics,the correlation coefficient between the average luminance characteristics and the geometric characteristics is calculated.It is found that the correlation coefficient between the mean luminance and the barycenter coordinate variance is 0.08759,and there is basically no correlation.Therefore,the barycenter variance and luminance characteristics of the flame image can be used as the basis for the identification of the sintering state,so as to reduce the input characteristic dimension.Several process parameters of sintering process are analyzed,the multi-parameter decision clustering of fuzzy C-means sintering state is analyzed based on the information of the average velocity variation trend and the flame image characteristics of tail section.Figure 44;Table 7;Reference 75...